Sistema embarcado para monitoramento de germinação de sementes em bandeja

Abstract

Agriculture is an area influenced by the Internet of Things, and machine learning, as well as computer vision have become tasks in this area that received the name of smart agriculture. Such an evolution encourages farmers to turn to technology to make their farms more efficient, automated, and better cost-benefit. Using cell trays for seedling production is one system widespread among producers. However, estimating the germination rate of seeds in trays requires analysis by an expert and becomes a time-consuming task, as well as natural resources such as water, which can often be wasted in the germination process and seedling formation of a crop. Considering these aspects, this work aims to present the development of an embedded system for monitoring seed germination in cell trays. The system recognizes, analyzes and presents the amount of germinated seeds and data from the sensors so that producers have control over the production yield of their crops and microclimatic information on the environment. The presented solution proved to be reliable and the germinated seed detection results were acceptable, with an accuracy of 82.76% and an f-score of 75.35%. The author's repository presents instructions for replicating the project and provides the dataset used. The microclimatic properties of the environment and the germination rate are relations that can be investigated in future works.